Bitmap/Flood fill: Difference between revisions

A flood fill is a way of filling an area using ''color banks'' to define the contained area or a ''target color'' which "determines" the area (the ''valley'' that can be flooded; Wikipedia uses the term ''target color''). It works almost like a water flooding from a point towards the banks (or: inside the valley): if there's a hole in the banks, the flood is not contained and all the image (or all the "connected valleys") get filled.

[[Image:Unfilledcirc.png|128px|thumb|right]]

'''Testing''': the basic algorithm is not suitable for ''truecolor'' images; a possible test image is the one shown on the right box; you can try to fill the white area, or the black inner circle.

=={{header|Action!}}==

In the following solution a simple implementation of queue has been used.

{{libheader|Action! Bitmap tools}}

RGB black,white,yellow,blue

{{out}}

[https://gitlab.com/amarok8bit/action-rosetta-code/-/raw/master/images/Flood_fill.png Screenshot from Atari 8-bit computer]

=={{header|Ada}}==

( Picture : in out Image;

From : Point;

end Flood_Fill;</syntaxhighlight>

The procedure has the following parameters. ''Picture'' is the image to change. ''From'' is the point to start at. ''Fill'' is the color to fill with. ''Replace'' is the color to replace. ''Distance'' defines the range of color around ''Replace'' to replace as well. The distance is defined as a maximum of the differences of stimuli. The following code snippet reads the test file, fills the area between two circles red, and writes the result:

File : File_Type;

begin

end;</syntaxhighlight>

=={{header|Applesoft BASIC}}==

110 COLOR= 12

120 X = 20:Y = 30: GOSUB 140"FLOOD FILL"

=== Recursive ===

This is limited to %StackSize% pixels.

CoordMode, Mouse

CoordMode, Pixel

=== Iterative ===

#SingleInstance, Force

DllCall("DeleteObject", UInt, hBrush)

}</syntaxhighlight>

=={{header|BBC BASIC}}==

BBC BASIC has a built-in flood fill statement, but to satisfy the terms of the task it is not used in this example.

GCOL 15

CIRCLE FILL 640, 512, 500

NEXT

ENDPROC</syntaxhighlight>

=={{header|C}}==

===Simple and complete example in C89===

* RosettaCode: Bitmap/Flood fill, language C, dialects C89, C99, C11.

*

===Second example ===

// http://commons.wikimedia.org/wiki/File:Julia_immediate_basin_1_3.png

The <code>sys/queue.h</code> is not POSIX. (See [[FIFO#C|FIFO]])

typedef struct {

color_component red, green, blue;

rgb_color_p rcolor);</syntaxhighlight>

typedef struct _ffill_node {

(Comments show changes to fill the white area instead of the black circle)

#include <stdlib.h>

#include "imglib.h"

return 0;

}</syntaxhighlight>

=={{header|C sharp|C#}}==

{{works with|C#|3.0}}

This implementation matches exact colours only. Since the example image has grey pixels around the edges of the circles, these will remain grey after the interiors are filled.

using System;

using System.Collections.Generic;

}

</syntaxhighlight>

=={{header|C++}}==

{{libheader|OpenCV}}

'''Interface'''

#define PROCESSING_FLOODFILLALGORITHM_H_

</syntaxhighlight>

'''Implementation'''

FloodFillAlgorithm::~FloodFillAlgorithm() {

</syntaxhighlight>

=={{header|D}}==

This version uses the bitmap module from the Bitmap Task, matches exact colours only, and is derived from the Go version (to avoid stack overflow because unlike Go the D stack is not segmented).

void floodFill(Color)(Image!Color img, in uint x, in uint y,

Using the image type from [[Basic bitmap storage#E]].

def matchColor := image[x, y]

def w := image.width()

[[File:Filledcirc-E.png|128px|thumb|right|Filled sample image]]Note that this does not make any attempt to smoothly fill 'banks' or have a tolerance; it matches exact colors only. This will fill the example image with red inside green, and there will be black/white fringes:

println("Read")

def i := readPPM(<import:java.io.makeFileInputStream>(<file:Unfilledcirc.ppm>))

println("Done")

}</syntaxhighlight>

=={{header|ERRE}}==

In "PC.LIB" library there is a FILL procedure that do the job, but the example program implements the algorithm in ERRE language using an iterative method. This program is taken from the distribution disk and works in 320x200 graphics.

PROGRAM MYFILL_DEMO

</syntaxhighlight>

Note: I haven't an "Upload files" item, so I can't show the resulting image!

=={{header|Euler Math Toolbox}}==

Using an emulated stack. EMT's recursive stack space is limited. For the notebook with images see [http://www.euler-math-toolbox.de/renegrothmann/Flood%20Fill.html this page].

>file="test.png";

>A=loadrgb(file); ...

>insrgb(B);

</syntaxhighlight>

=={{header|FBSL}}==

'''Using pure FBSL's built-in graphics functions:'''

#DEFINE WM_CLOSE 16

END SUB</syntaxhighlight>

'''Output:''' [[File:FBSLFlood.PNG]]

=={{header|Forth}}==

This simple recursive algorithm uses routines from [[Basic bitmap storage]].

: 3dup third third third ;

: 4dup 2over 2over ;

then

r> drop ;</syntaxhighlight>

=={{header|Fortran}}==

{{works with|Fortran|90 and later}}

Here the ''target color'' paradigm is used. Again the <code>matchdistance</code> parameter can be tuned to ignore small differences that could come because of antialiasing.

use RCImageBasic

use RCImagePrimitive

Usage example excerpt (which on the test image will fill with green the inner black circle):

=={{header|FreeBASIC}}==

{{trans|BBC BASIC}}

' compile with: fbc -s console

If InKey <> "" OrElse InKey = Chr(255) + "k" Then End

Loop</syntaxhighlight>

=={{header|Go}}==

An addition to code from the bitmap task:

func (b *Bitmap) Flood(x, y int, repl Pixel) {

And a test program. Works with code from read ppm and write ppm to pipe tasks. For input, it uses a version of the test file converted by the Go solution to "Read an image through a pipe". For output it uses the trick from "PPM conversion through a pipe" to write the .png suitable for uploading to RC.

[[File:Go_flood.png|right]]

import (

}

}</syntaxhighlight>

=={{header|Haskell}}==

This code uses the Bitmap and Bitmap.RGB modules defined [[Bitmap#Haskell|here]].

import Data.STRef

import Control.Monad

scanWhileX b st p oldC newC (w, h) (Pixel (x, y + 1))

</syntaxhighlight>

=={{header|HicEst}}==

HicEst color fill is via the [http://www.HicEst.com/DeCoRation.htm decoration option of WRITE()]

WRITE(WIN=wh, DeCoRation="EL=14, BC=14") ! color 14 == bright yellow

WINDOW(Kill=wh)</syntaxhighlight>

=={{header|J}}==

'''Solution:'''<br>

Uses <code>getPixels</code> and <code>setPixels</code> from [[Basic bitmap storage#J|Basic bitmap storage]].

NB. ref: http://www.jsoftware.com/pipermail/general/2005-August/023886.html

findcontig=: (|."1@|:@:>. (* * 1&(|.!.0)))^:4^:_@(* >:@i.@$)

The following draws the same image as for the [[Flood fill#Tcl|Tcl example image]] below.<br>

Uses definitions from [[Basic bitmap storage#J|Basic bitmap storage]], [[Bresenham's line algorithm#J|Bresenham's line algorithm]] and [[Midpoint circle algorithm#J|Midpoint circle algorithm]].

myimg=: white makeRGB 50 70

lines=: _2]\^:2 ] 0 0 25 0 , 25 0 25 35 , 25 35 0 35 , 0 35 0 0

'''Alternative findcontig:'''<br>

The following alternative version of <code>findcontig</code> is less concise but is leaner, faster, works for n-dimensions and is not restricted to numerical arrays.

eq=:[:}:"1 [:($$[:([:+/\1:,}:~:}.),) ,&_"1 NB. equal numbers for atoms of y connected in first direction

eq_nd=: i.@#@$(<"0@[([:, |:^:_1"0 _)&> [:EQ&.> <@|:"0 _)] NB. n-dimensional eq, gives an #@$,*/@$ shaped matrix

findcontig_nd=: 3 : '($y)${. ([:({.,~}:) ([ repl cnnct)/\.)^:([:+./@(~:/)2&{.)^:_ (,{.) eq_nd (i.~ ~.@,) y'</syntaxhighlight>

=={{header|Java}}==

Input is the image, the starting node (x, y), the target color we want to fill, and the replacement color that will replace the target color. It implements a 4-way flood fill algorithm. For large images, the performance can be improved by drawing the scanlines instead of setting each pixel to the replacement color, or by working directly on the databuffer.

import java.awt.Point;

import java.awt.image.BufferedImage;

}</syntaxhighlight>

And here is an example of how to replace the white color with red from the sample image (with starting node (50, 50)):

import java.awt.Color;

import java.awt.Point;

}

}</syntaxhighlight>

=={{header|Julia}}==

{{works with|Julia|0.6}}

Inspired to [[#Python | Python]] version.

function floodfill!(img::Matrix{<:Color}, initnode::CartesianIndex{2}, target::Color, replace::Color)

floodfill!(img, CartesianIndex(100, 100), Gray(false), Gray(true))

save("data/filledcircle.png", img)</syntaxhighlight>

=={{header|Kotlin}}==

{{trans|Java}}

import java.awt.Color

JOptionPane.showMessageDialog(null, JLabel(ImageIcon(image)), title, JOptionPane.PLAIN_MESSAGE)

}</syntaxhighlight>

=={{header|Liberty BASIC}}==

NoMainWin

WindowWidth = 267.5

End If

End Function</syntaxhighlight>

=={{header|Lingo}}==

Lingo has built-in flood fill for image objects, so a custom implementation would be pointless:

=={{header|Lua}}==

Uses Bitmap class [[Bitmap#Lua|here]], with an RGB tuple pixel representation, then extending..

Preprocess with ImageMagick to simplify loading:

Some rudimentary PPM support:

local fp = io.open( filename, "rb" )

if fp == nil then return false end

end</syntaxhighlight>

The task itself:

local b = self:get(x, y)

if not b then return end

end</syntaxhighlight>

Demo:

bitmap:loadPPM("unfilledcirc.ppm")

bitmap:floodfill( 1, 1, { 255,0,0 }) -- fill exterior (except bottom right) with red

bitmap:floodfill( 100, 100, { 0,0,255 })-- fill smaller circle with blue

bitmap:savePPM("filledcirc.ppm")</syntaxhighlight>

=={{header|Mathematica}} / {{header|Wolfram Language}}==

RegionBinarize[img, Image[SparseArray[pos -> 1, ImageDimensions[img]]], tol];

floodFill[img_Image, pos_List, tol_Real, color_List] :=

Import the test image and fill the region containing the pixel at coordinate 100,100 with red (RGB 100%,0%,0%) using a tolerance of 1%

<pre>floodFill[Import["http://rosettacode.org/mw/images/0/0f/Unfilledcirc.png"], {100, 100}, 0.01, {1, 0, 0}]</pre>

=={{header|Nim}}==

{{Trans|Python}}

proc floodFill*(img: Image; initPoint: Point; targetColor, replaceColor: Color) =

img.floodFill((30, 122), White, color(255, 0, 0))

img.writePPM("Unfilledcirc_red.ppm")</syntaxhighlight>

=={{header|OCaml}}==

{{Trans|C}}

let floodFill ~img (i, j) newColor =

let oldColor = get_pixel ~img ~pt:(i, j) in

in

aux (i, j)</syntaxhighlight>

=={{header|Pascal}}==

{{trans|C#}}

program FloodFillTest;

end.

</syntaxhighlight>

=={{header|Perl}}==

The <tt>fill</tt> of the Perl package Image::Imlib2 is a flood fill (so the documentatin of Image::Imlib2 says). The target colour is the one of the starting point pixel; the color set with <tt>set_color</tt> is the fill colour.

use strict;

A homemade implementation can be:

use Image::Imlib2;

This fills better than the Image::Imlib2 <tt>fill</tt> function the inner circle, since because of JPG compression and thanks to the <tt>$distparameter</tt>, it "sees" as black also pixel that are no more exactly black.

=={{header|Phix}}==

{{Trans|Go}}

Requires read_ppm() from [[Bitmap/Read_a_PPM_file#Phix|Read a PPM file]], write_ppm() from [[Bitmap/Write_a_PPM_file#Phix|Write a PPM file]]. <br>

Uses the output of [[Bitmap/Midpoint_circle_algorithm#Phix|Midpoint circle algorithm]] (Circle.ppm), results may be verified with demo\rosetta\viewppm.exw

include ppm.e -- blue, green, read_ppm(), write_ppm() (covers above requirements)

img = FloodFill(img, 10, 10, green)

write_ppm("FloodOut.ppm",img)</syntaxhighlight>

=={{header|PicoLisp}}==

Using the format of [[Bitmap#PicoLisp|Bitmap]], a minimal recursive solution:

(let Target (get Ppm Y X)

(recur (X Y)

(ppmFloodFill (ppmRead "Unfilledcirc.ppm") 192 128 (255 0 0))

"Filledcirc.ppm" )</pre>

=={{header|PL/I}}==

declare (x, y) fixed binary;

declare fill_color bit (24) aligned;

The following PL/I statements change the color of the white area

of the sample image to red, and the central orb to green.

/* Fill the white area of the suggested image with red color. */

area_color = (24)'1'b;

call fill (125, 125, '000000001111111100000000'b );

</syntaxhighlight>

=={{header|Processing}}==

import java.util.Queue;

import java.util.LinkedList;

==={{header|Processing Python mode}}===

image_file = "image.png"

img.pixels[pixel_position(x, y)] = fill_color

return True</syntaxhighlight>

=={{header|PureBasic}}==

=== built-in ===

; Fills an Area in red</syntaxhighlight>

=== Iterative ===

old_color = Point(x,y)

NewList stack.POINT()

Until Event = #PB_Event_CloseWindow

EndIf</syntaxhighlight>

=={{header|Python}}==

import Image

def FloodFill( fileName, initNode, targetColor, replaceColor ):

===Usage example===

# "FloodFillClean.png" is name of input file

# [55,55] the x,y coordinate where fill starts

img.save( "Filled.png" )

</syntaxhighlight>

=={{header|R}}==

'''Stack-based recursive version'''

library(png)

img <- readPNG("Unfilledcirc.png")

</syntaxhighlight>

'''Queue-based version (Forest Fire algorithm)'''

library(png)

img <- readPNG("Unfilledcirc.png")

image(M, col = c(1, 0, 2, 3))

</syntaxhighlight>

=={{header|Racket}}==

#lang racket

bm

</syntaxhighlight>

=={{header|Raku}}==

(formerly Perl 6)

Using bits and pieces from various other bitmap tasks.

class Bitmap {

has Int ($.width, $.height);

See output image [https://github.com/thundergnat/rc/blob/master/img/Bitmap-flood-perl6.png Bitmap-flood-perl6 ] (offsite image file, converted to PNG for ease of viewing)

=={{header|REXX}}==

{{trans|PL/I}}

black= '000000000000000000000000'b /*define the black color (using bits).*/

red = '000000000000000011111111'b /* " " red " " " */

@: parse arg $x,$y; return image.$x.$y /*return with color of the X,Y pixel.*/</syntaxhighlight>

<br><br>

=={{header|Ruby}}==

Uses [[Raster graphics operations/Ruby]]

require_relative 'raster_graphics'

JRubyArt is a port of Processing to the ruby language

Pixel = Struct.new(:x, :y)

end

</syntaxhighlight>

=={{header|Rust}}==

/* Naive Rust implementation of RosettaCode's Bitmap/Flood fill excercise.

*

}</syntaxhighlight>

=={{header|Scala}}==

See [[Basic_bitmap_storage#Scala|Basic Bitmap Storage]] for RgbBitmap class.

import scala.collection.mutable

}

}</syntaxhighlight>

=={{header|Standard ML}}==

This implementation is imperative, updating the pixels of the image as it goes.

data structures instead.

* fundamental would change if we used more colors. *)

datatype color = Black | White

(* Fill the image with black starting at the center. *)

val () = fill test Black (3,3)</syntaxhighlight>

=={{header|Tcl}}==

{{libheader|Tk}}

{{tcllib|struct::queue}}

Using code from [[Basic bitmap storage#Tcl|Basic bitmap storage]], [[Bresenham's line algorithm#Tcl|Bresenham's line algorithm]] and [[Midpoint circle algorithm#Tcl|Midpoint circle algorithm]]

package require Tk

package require struct::queue

[[Image:Tcl_flood_fill.png]]

=={{header|Wren}}==

{{libheader|DOME}}

When the up arrow is pressed, the red square changes to blue and when the down arrow is pressed the blue square turns back to red.

import "dome" for Window

import "input" for Keyboard

var Game = Bitmap.new("Bitmap - flood fill", 600)</syntaxhighlight>

=={{header|XPL0}}==

[[File:FloodXPL0.gif|right|Output]]

proc Flood(X, Y, C, C0); \Fill an area of color C0 with color C

SetVid(3); \restore normal text mode

]</syntaxhighlight>

=={{header|zkl}}==

[[file:Flood_before.zkl.jpg|right]][[file:Flood.zkl.jpg|right]]

Uses the PPM class from http://rosettacode.org/wiki/Bitmap/Bresenham%27s_line_algorithm#zkl

targ,h,w:=pixmap[x,y], pixmap.h,pixmap.w;

stack:=List(T(x,y));

}

}</syntaxhighlight>

pixmap.circle(101,200,100,0); pixmap.circle(75,100,25,0);

pixmap.writeJPGFile("flood.zkl.jpg");</syntaxhighlight>

{{omit from|Computer/zero Assembly|this language doesn't support video output and only has 32 bytes of RAM}}

{{omit from|Lotus 123 Macro Scripting}}

{{omit from|PARI/GP}}